Talk
in
Workshop: Machine Learning for Audio Signal Processing (ML4Audio)

Singing Voice Separation using Generative Adversarial Networks

Hyeong-Seok Choi · Kyogu Lee

2017 Talk
in
Workshop: Machine Learning for Audio Signal Processing (ML4Audio)

Project Page

Abstract

(+Ju-heon Lee) In this paper, we propose a novel approach extending Wasserstein generative adversarial networks (GANs) [3] to separate singing voice from the mixture signal. We used the mixture signal as a condition to generate singing voices and applied the U-net style network for the stable training of the model. Experiments with the DSD100 dataset show the promising results with the potential of using the GANs for music source separation.

Chat is not available.